What's Changed
- Make it easier to debug nested tensors. by @comfyanonymous in #15383
- Update workflow templates to v0.11.37 by @comfyui-wiki in #15415
- Minimum officially supported pytorch is now 2.7 by @comfyanonymous in #15413
- fix: make Create Layered Image discoverable and its flags self-explanatory by @jtydhr88 in #15429
- Fix upscale models breaking on non dynamic vram low vram. by @comfyanonymous in #15437
- Add previewable_outputs_count to /api/jobs (backend half of Media Assets badge fix) by @claude[bot] in #15148
- Extend ER-SDE noise scaler by scaling h(t) by @chaObserv in #15428
- Make a context manager for cast_bias_weight and use it. by @blepping in #14750
- Optimize MiniMax-H3 VAE by @kijai in #15446
- [Partner Nodes] feat(Qwen): add Qwen-Image 3.0 Pro by @bigcat88 in #15327
- Add tags_all / tags_any / tags_none tag filters to the assets list API by @synap5e in #15332
- Fix VAEDecodeTiled crash on NestedTensor latents (MiniMax H3) by @chelsealong in #15477
- Make cu130 warning more visible. by @comfyanonymous in #15463
- Implement comfy kitchen attention. by @comfyanonymous in #15479
- Fix peak memory issue with H3. by @comfyanonymous in #15486
- Add support for LTX 2.5 by @alexisrolland in #15499
- [Partner Nodes] feat(LTX): add new nodes for model version 2.5 by @bigcat88 in #15501
- Fix for broken tiled audio decode. by @comfyanonymous in #15502
- Update workflow templates to v0.11.39 by @comfyui-wiki in #15504
- [Partner Nodes] feat(Grok): add Grok-Imagine-Image-2.0 model by @bigcat88 in #15496
- Fix some clip vision regression. by @comfyanonymous in #15506
- Don't depend on transformers for mistral and llama tokenizers. by @comfyanonymous in #15503
- Remove potentially problematic process_tokens method. by @comfyanonymous in #15507
New Contributors
- @claude[bot] made their first contribution in #15148
- @chelsealong made their first contribution in #15477
Full Changelog: v0.31.0...v0.32.0